32 I

Mark D. Flood

Mark D. Flood is an economist at the Federal Reserve Bank of

St. Louis. David I-I. Kelly provided research assistance.

I
An Introduction to Complete
Markets

I
HE PAST TWO DECADES have seen a pro-
liferation of new and often complex financial ~yHm~ Saths Starts SeIiin~New
securities and commodity contracts iii the mar- W~1~F
nts Rettin$ Yen vs. Mark
ketplace. Flipping through the financial pages of
the newspaper, one finds, for example, that an
____________
investor can purchase the right to buy (at a fixed ~ WALL TRFZT JOURN Staff Itcpener
price on a set future date) a futures contract on ~ f~f~J4~a tis ti lipgsn
lb Slion
10-year U.S. Treasury notes. Alternatively, one ant Sh,So i~ a w lit q b t di
reads of a contract that pays off various dollar S°~ 3/eriiWflStftti~lbe against Lie tnatslnerk
amounts depending on the level of the mark- d~’~SbColdtnanh thqy be
yen exchange rate at the end of October 1992 ~ of the coat
(see shaded insert at right). Are such arcana niuni ~ ompan nap o th off t-ing F
practically useful? The theory of complete & 0 0 an W at nold I
- A carding to e optic r 4 e are li-st
markets—an important element of modern ~ e . thu ghsitch

theoretical economics—can provide some arrantutsateb oldra ope Thew ant tartedtrading
insight.’ yasterday is the e’ V Ic Stoc oh nga
01 ra F options ga hotter thengh hut b
A complete system of markets is one in which itt o6liga e oh er II securt,e tnt xi ies arid oth
there is a market for every good. This simple ‘ *t tP~i thpnceonorhe Fe athda Tb
- - - ap’ n i mor ompli dbe a ext nit’ olderstob
statement conceals the significance of the con- ~ ~ ~i~oforean sire ivithap of
cept, however, by failing to specify what is i dol
meant by a “good.” By carefully defining “good” Th Walt t oxtrualpe babe a 41* 1 of cit rate
to include the date and environment in which a
- - -
between nosi cur i
sterdavw 8 68 yantobu iu amp ooemathe
Ic to g to am
commodity is consumed, economists are able to Thew ran nfl anitive tortoapavni nfl U. d fist
consider consumption, production and invest- numb v atopu haseanent tIc drop el iv So 20
ment choices in a multiperiod, uncertain world. T ts alueo thewar ntssc tutu edfrorn foi-nrnla
- p Li warrant p o pecitis Each vat aittun ally’~ill
Moreover, they can do so using largely the same ~ $ ~ ~ aria e pires o~ o 80
utility theory originally developed to analyze ~l~nof tIe l000wa i Wilinu Liar
timeless certainty. In particular, state-preference iii ltpw tiontb ea e e can tyle
i-ran an
theory, which was developed to analyze the Li do yb sirs ay ReM wilini obeabi
-. - Se ti fit 8’ floart oh rxvsto
completeness of a system of markets, is a pow-

‘To be correct, one should refer to a “complete system of “complete market.” These three terms are used inter-
markets.” This phrase can be unwieldy at times, however, changeably here (see glossary).
and it is usually abbreviated to “complete markets” or

FEDERAL RESERVE BANK OF St LOUIS a

F
A Glossary for Complete Markets

.\rbitrage .~ collection of pavoll n 101’s ~\ liii it~ of p~i~

oIl au oss stales see ~ilso
zero net osI Vt 110’”’ net P’~.)oils ill ciII out— .‘-pPtIJIHIP).
rOflW~3! P non—negative ~mci riot all iero In
othei t~oici a set ol tr ansacLions tliaL re— Inc oniplete market: ~t s~stem of niai’kets

stills in getting something liii- nuIliiri~,. Vt hit h is not I omplc’te

II A rron -lJc’brcu serur itt

seem-itt.
S~tnieas pun-

Ret: \ ontrac r it hose payoff’, are a coristani

l.iiu’ax t’oiuhirti tioxi A (?(‘toi’ pi ixiric Cl by
at lii ing (or sub Li ac-tn gi t~IcmenI hi ele-
nieiit, ant sralar multiples of hi o or 111011’
itiotint in all otitcoiiits that iiI’P t’lt’itieiits otlici ‘— (‘(‘lois.
ol a specified ci mt incl zero in all othex -

otitconies. I-ui etaniple ,t t oiilrat’t thit I meat lv independent

- - -
\ set ot ec-tors is liii—
i’s ott S5 in t he N exit - \ I it ins alit] ZPIO
• r 1> inch’; jen dent ii 110 t et Lot in the Wi
-
citlirrit se b a bet can he c’xpr’esWcl as a linear tonil,mnatioii
ol liii at hPr \ ttU ii s Iii tin’ 5( ‘I
all optu 1fF -‘ contract git log 11w holdem the
right to IJLJ~ 1 ~tti tIc tihW good at a ; )i’eS~W- 0 fi ion: See call opt ion and put option
t.rfini luture date trw a presp cit ied pi it e
(see -il Sc) put oj ,th~ - ( )u L( 1 me ~ coin~ ilel e spec if icat i Hi il the
i-eat iza twos of ~tlI i eit’i ant a riables over
rimpie It’ max ket: A sys I ciii cit i naik e Is fl tin en Li i-v n-in ant tin ir horizon see a l-,o

“ vi I en
ut Ii good, it herL di
vi P1% agetu is I_Pc Ic tolii t’.~c
Li I) tlt intl
change
rec liv (J vl~
with evt-rt other’ agetit \lsit c’aJIetl ~itoni— I’aic’to—optinal allocation An alIo( anon of

plele ~ St ets~ odds:

it sU 111thi rig
ni citBelt kets.
odds on some col- ii that
crononi
slit ic no I ruclit mdciamong
resources lip n iade bet Ic
-ii can individuals
Ic’i -t 11)11 of events that al-I’ inverseit i-cia ted ~fi Wit Ii out iiia ki tig some cit her i idi it] ual
to a probahi liii iiiea 5U I’(’ cit er t I lose Pt CI) t S Vt Disc ol f

according
where D~is the odds
to the torniula ()~evei
on the — (1.it PU-I)
e a —rid1 PatrosenL
off vector-
the a mount
.\ vet-I inpaid
ichose c’lenients
of eat I ep-
h good in
Pt e) is Li ie’ probability ot e oc (Li! ring each ott tc’onie tindci a partic-Lila r corfl rat I -

F flit-lent a flotation: Same as Parita-optirnal

,illoriittori Pi-i nut it e sen it-i t} - Same as pure se’ ui-itt’
lien t: ~. ci illection of out comes. lot eta tnple Pu re security At’ on Liar L thaI payc C itt tine
‘ui it iris’ is an ci nit that includes the unit (usualk one dollar-) in a partic-ulai out-
. rarlkmgs \.l- 14 and X1-( -- F lsee also come and nothing in all other outcomes
outran it’) -
Put option: c-onti act git ing the holder the
-‘u

Futures cot it mac L: ‘u s Landa clued con Ii act to right to sell a ~3 rt iculam good at a prt-spe( I-
purc base a tornmoditv, whit h IS to he de- ill future ilit L P Iou’ a pres pet-if led inre
lit ert’d at a siwci iii date in the f u I tue (ste a so (all option)

I Gei ici-al eq u hbriu in - A feasible cci nihi nat on

o I pr-ic c s, -oiistinu pt ion c lioi n’s at ud pro-
citic I ion c’ inn es suc-i i that then’ is no cx-
Red ii: tdarut Iii t bet n I uose p~ttoff vec LOT
- - A

tail he cons Er uct ed a. a lii iea.i corn hi nat on

cess supph 01 demand for ant good. of the payoff vet ton at oWcr a~ai able
bets
Iiecigc - lo a rca nge a ~‘omhina Iioi u of con Li-ac Is
5 Lit-li I hat iou pavc Clhs for one c’ont rat I in a Scala r m ul tip] t-: .\ vecto F P oduc ed I it sc t]i rig

gic en ‘taLc-
another
result is contract
arc Iolio
a pnr’t oftinselwith
the high
hi same pat
ed offs
i’educslate. br
taxIRe
iabil- another
s,unevec-tor
nultipl~ing
11w eatnunnbt
real ox clown
Iiupeleiiii’rut i
of the lal tem bc . In

I
S
 34
I
U

1$
S20iL t j~rttN
j~
bfItttr~
~jdrrow~

N
good\wfl
ngTh
stat
aa\

Saa~ aztO
hrghusett
N
txxl
I
\\$ ~ spa ;seplS4~ic\ ~Mid~ ~ utrN that~aysoffi~NS~
Nphecd tS~ mail. ~Qrdifbrtgocd)underdlfft
that \l?0Mt~t\pi ~eiiQ ~othwqbt
narcrnn a
I
flQf\\ N

~IiO& tat\\N\~ \\ St~ titID t\IanIISaU%C tte en

~p~cuJat Toj$r~hase s4 a goad ~ Sta Ihe wo 14 Saffi as oflm
~táthng ~* ~‘l~bi~ lag, w~ %ecao r ad fleettanuf tmlbe
th~mbS4i
pled *
pr if en
tt
\i
lb an
~tart~ ~kobø pmsenttl pota~N
I
erful tool with which to study behavior under
uncertainty.
in a particular model. Second, although the no-
tion of market completeness appears most often
I
The purpose of this paper is to introduce the in theoretical discussions, the ideas involved can
non-specialist to the basic theory of complete
markets, providing the reader with an insight
also be applied to more realistic problems. For
example, in the state-preference context, mar-
kets for so-called “derivative” securities fu- —
I
into the nature of markets and recent financial
innovations in particular. The paper first intro-
duces the major concepts of the state-prefer-
ence approach to uncertainty, illustrating them
tures and options add value by providing
—

investors with flexibility in fashioning their

with a parimutuel gambling example. In this
framework, the notion of completeness arises
naturally as the extreme case in which bettors
have the greatest range of opportunities to bet
on the outcome of a race. The terminology is
then transferred to an economic context) where
again, complete markets provide consumers,
producers and investors the most flexibility in
allocating payoffs and planning for uncertain
contingencies. Particular attention is given to
the markets for futures and options.
Such securities are shown to improve the effi-
ciency of the marketplace, a result with implica-
tions for regulatory policy.
In the real world, systems of markets are not
complete, as we shall seeThe notion of com-
pleteness, however, is of interest for two rea-
sons. First, it serves as a theoretical benchmark,
relative to which incompleteness can be assess-
ed; such a comparison might, for example, sug-
gest whether incompleteness implies inefficiency
portfolios; thus, they make systems of markets
II
less incomplete. The popularity of such securi-
ties can thus be explained from a theoretical
perspective that incorporates complete markets.
in some cases, the theory can even suggest
new markets that would alleviate existing
incompleteness.
THE THEORETICAL APPARATUS
The tools and results of the theory of com-
plete markets represent one the most significant
developments of theoretical economics in this
2
century. At the same time, the concepts enihed-
ded in the theory are very general and have
been used in many other economic contexts.
Thus, our first task is to explore the basic struc-
ture of state-preference theory. We do this with
a simple gambling example, because it involves
a well-defined and relatively small collection
of outcomes and payoffs in an uncertain

2
environment.
3
I
The theory can be traced to the work of Arrow (1964), counterpart, the theory of incomplete markets. The mitera-
Debreu (1959), Arrow & Debreu (1954) and Mckenzie
(1954) in the mid-1950s. Both Arrow and Debreu were
later awarded the Nobel Memorial Prize in Economics
(Arrow in 1972, Debreu in 1983), largely for their work in
ture applying state-preference theory and the theory of
complete markets is lengthy; for a partial list, consult the
references in Radner (1982) and Debreu (1982).
1
Previous articles have recognized the connection between
developing the theory of complete markets and applying it
to the problem of general equilibrium. The theory is often
cited in the guise of its two most common avatars, the
Arrow-Debreu general equilibrium theory and state-
gambling and economic activity, especially financial
markets. See, for example, Gabriel & Marsden (1990) or
Asch, Malkiel & Quandt (1984), and the references therein. I
preference theory; it also often appears as its implicit

I
FEDERAL RESERVE BANK OF St LOUIS
S
 35
II I
State-Claims Defined T-M-C, T-C-M, M-T-C, M-C-T, C-T-M, C-M-T6 Al-

I One dominant theme of state-preference

theory and complete markets is uncertainty,
though we may have definite opinions at the
start of the day, we cannot know the state of
the world for sure until the race has been run.
I State-preference theory incorporates uncertainty
by defining outcomes, or potential future states,
An event is a collection of one or more states,
~ only one of which will ultimately be realized,
‘Fhus, for example, “Mastercharger wins the
-
race” is an event, It includes two states, M-T-C
The theory has been fruitfully applied in many and M-C-T, both of which are consistent with
I areas of economics, especially in the study of
financial markets. For now, however, consider
an imaginary racetrack called Portfolio Downs.
the stated criterion,

All bets are placed on events. If the state of

We are bettors at Portfolio Downs, hoping to
win enough for an early retirement. To make
our life simpler, the track’s management has
done away with the tote board.~Instead, all
bets are placed at fixed odds with the track’s of-
ficial bookmaker. This allows us to confine our
uncertainty to the race itself, without worrying
that the posted odds might shift after we’ve
placed a bet. Furthermore, to keep the number
of contingencies to a minimum, only one race
will be run today.
Our first assignment at the track is to define
the set of relevant states and payoffs. A state
the world that ultimately occurs is an element
of the event, then the bet pays off at the fixed
odds; otherwise the bet pays nothing. Bets are a
special type of state-contingent claim (or simply
state claim). More generally, a state claim is a
contract that pays off differing amounts—per-
haps even different goods—under different
states of the world.
A state claim can be represented as a payoff
vector with one element for each state of the
tvorld. The notion of a payoff vector, central to
the theory of complete markets, is stated in
terms of the mathematics of vectors, linear alge-

I
~
of the world is defined as a complete specifica-
tion of the values of all relevant variables over
the entire relevant time horizon. A state of the
bra! In our example, a $2 bet on Mastercharger
to win that pays off at 4-to-i odds can be repre-
sented by the vector: (O.O,$iO,$iO,O,0), where

I world is also called an outcome, or simply a

state. In an economic system, the relevant van-
ables might include the structure of the tax
code. global weather patterns, infant mortality
the positions in the vector are in the same
order as the states listed above. Alternatively, a
$2 trifecta bet on the ranking C-T-M that pays
off at 3-to-i odds has the payoff vector:

I rates, scientific discoveries, etc. ‘I’he relevant

time horizon might be stated in periods as long
(O,O,O,O,$8,O).

In state-preference theory, a market is

as a decade or as short as a second; it might cx-

I tend into both the past and the future for a few
hours or many centuries.
At the racetrack, however, matters are much
equivalent to a payoff vector: a market repre-
sents the ability to exchange goods or payoffs.
At the racetrack, we exchange pre-race dollars
for a state-contingent bundle of post-race dol-

I simpler: a state of the world is a complete list-

ing of the finishing position for every horse in
5
today’s race, For example, if there are only
lars. Some exchanges are available directly: we
can exchange $2 pre-race for the post-race vec-
tor (O,O,$1O,$iO,O,O). Other exchanges can be

I thre~horses in the race:

1. Tricky Bond (T)
constructed indirectly: we might exchange $4
pre-race for the post-race vector (O,O,$iO,$iO,
$4,$4) by placing two separate $2 bets.
2. Mastercharger (M)

I ~ Charge Me Interest (C)

then there are six possible states of the world,

The key to the theory of complete markets is
to deal with such combinations in a systematic
which we can write in win-place-show order: fashion. A system of markets is complete when

I 4
Some basic racing and betting terminology is defined in
6
ln general, the number of permutations is given by the fac-
tonal function. E.g., 3! = 3-2-1 = 6.
7
5 See the shaded insert on pp. 37-38 for a quick introduc-
the world
Noteshaded
that
would insert
if there were
ona page
involve more36.thanlisting
complete one of
race, a state inof all
all horses the tion to linear algebra.
races. In other words, a single state of the world describ-
ing all the day’s races would be ultimately realized.

MARCF,’APRtL 1991
S
 k\C\ ~ ~ N ~NNN N N

Parisflhetuettia% NN N

Seitost Clossary
N N N N N

/ / ~ fr~ f~ N
N
N

W
N/N
NJIS
~ ‘weanig~ st~ke and
\Ntt~
nwtuS n~
tslbe ~tS zScNbarn ~ 1) 47 Doublehorsesbetinthato desg$tted
Wunun peaftes tlwSe
tWO /

ttg&4ttifl ddtend~iaoffsIntb N~ th~WtIe~

ax ommoidy used a Amencn rage- pa a f
~S~~’° I?0~t~USeSNa E Satáas ecth
kniak ~ PS
\thtp~~nt~mologykEthesaruewsbath / C) d-Th atwof\netpDaf%t toa~nottu
zuetiab. a. paàrtutuS stern bet a e w red I r a winning trek
pppped together in JIOQIS WneapendIpg to N N

~7pet~fl.\ p how qurn Pan weger~ ~sequene twoormore\

eø~tthStwtr~ted~partdy T~h. ~bs \snbzhancoptingenct
N~ tqp ai~II ~ tI*N*W u ads- and 4n latch the full proceeds
rethSnm~podt~ ~ngth Np~i~a4 81 her bet e i~ gored oathe St INN

~mnmg bott N ~ ml anslmt be the s~q~ten a F stiple N daily dent

ad rèoofprohi N blebetàaparia ~~“eN

N
N j
/ /P
tar 2t%SiØ
tiler belt
tWtnflIJt
niakut tt$alt$& The
$2 hEft t ~ Pefecta
bq
A Sstat
set to order ~ee aiao~4Vi 1)
N

1
jtt/ N ~ Nm Ondana ace to aN

N / N ~nt
jh flOQLhIpa4iS~ be a$ndba ~ N

~jhe odds ORN

ta ~tS N Ii ~ ‘qrp r late~
N~ v4,—Nsm~j— trite q~ the N N N
N hone pFtbøj3tIQ,E ~N~i O~tS~N q ni Abetthetsp N

// / poW 43db istlieanioN ~ N N btd *otapmc ~ N

NIN Stwa’betoohotsebodds erdersee — NNN NNN

nzatbotOtLbostandflu ~ N NNNN N N\N N\/

/N\ ~rahS fbdtiplaccdnp N~ ho

~ ebj$Ithr~~tlu \NN N N\~~ tie d1OWNN~\NN dbi~ cxvii
~angmumnii ~~nX~fpWbS4N~1n ~ N\ tLa~wtn P ~ N:

w
teSs tit*~r the
\~°~N.N t4iaPL: ~ u4h It jjppba~
TM~tfrnn~th betndgpàalthat
tfl~ni~ythe racetiack ~ I I
~k Nt hwdtpa~off~Th
at! iaNbor £ \N N

her onto ~ q~n ~TotalIz toni ot&boer ~A a stem oN reptit

W\N ft~N\ \P JtttiWty N NN tingahor a urrent~ettngoddsimjib&\
lie ~piebet. peedy horse to WITh N jn the mu tie! pool N

the It! \jtoal~~

the finalodds on~thatho so to N N N
N ~ cedtheoddsonthesamehere rlfitct Abétthats~e& esth top~tw
taNw~n ~ N the pIece bet r a
tbur~4I inshers> ~ordcr NW N

thenllieMnbt Althoughtbrs IearI7hn~N

p ies tnadnsistettt ad it a and occasion
a*doesoccur
we can arrange a portfolio with any conceivable
payoff vector. We may not want certain payoff
vectors, and even among tho e we do find de
sirable, our decision on which portfolio
ultimately to arrange will depend both on their
prices and our resources, These i sues of desir-
ability and affordability are secondary for th
notion of completenes however. The important
N Win To caine in ft t in a AIsa be
mba a. pethedharsewifiwmn.
characte istic of a complete system of markets
is that, without a wealth constraint, any con-
ceivable payoff vector can be arranged. In
terms of hnear algebra, a complete ystem of
markets is one in which the set of available be s
contains enough linearly independent payoff
vectors to span the space of all conceivable
, payoff vectors.

S
I p a — a a a aa S S S —SH--—S a

1~
C
3,.
1,

r
~0
(0
 38
on each of tire elements.’ (Ji-aplth-all~ this
i’esulis iii tlit’ Iengtheriiig 0!’ ~hoL’tening til
11w original ~‘ec’toi’. n’ithotit changing its
dir’ec’tioii, In our secLirlues example if n C are
afloned to purchase fr’ac’iiona] amounts of
the seen t-it les for’ our port ioliu. then ~ ~ ~
obtain a pavoft vector that ]ies am-n he’~’ein
he pt u-pie-shaded wed “e.
sc’,iIui’ tIltdti )ItCJtiOtl 01 a cc-tot’ also n nihs
1
U Fth negnti~t’ tliit!lllt’r sit ne moltiph ~ ec—
tot’ fl (I tlc’gat R C’ tlLttlil)c’t’, I hi’ i’i~tltit;g ~
11 tilt’ graph lies tlOllg t]ir’ “aim’ line as miii’
or’igirial htit points ill the tippo~iti clii’i’c’tiuri.
tot’ out’ st’CLlt’ities ex unple hi’, is t hr c’qttr~
1
aunt of a ~shor’!sale the salt’ tit a hor-t’m~id
set twit~- Shot t salt’s tilt genc’ralh made n flil
the itttentiotl OF t’efiLirrhasnlg tlii’ security
later at a I!lopl’fLtll\ I bin et pm-itt’. It no ‘~eiIa
sec’nr’it~ shtu’t, U C’ art’ t’espc)tlsi ,lt’ for t’eitn
hui’sitig tin’ ieticlt’t’ tic’ amount ol tin’ sec’tu’i-
~ pa~ciii n hen lit’ state nt he nrn’ld i’~
in etled —— lic’rn e Wi’ negati~C’ jnivolfs \hithe-
tnatii’d1~, tins r~the sjnti’- as ‘
ZldditlQ
‘
a ni-ga-
- Vt till tflOI’(’ thiti Ln (I states ot the world, I Ill’
tnt’ sec-ut’iR to out’ pot’ttolio: P —‘ P = P —
~ ‘ /
- I’,) (a’aplticalh_ nt’ add the at’t’o\\” in the
saulli’ Vt JV. making stile that Ut’ tist’ the I I’,,)
‘~c’i tot in I lie tclclition. It n r’ itt’ aIlnt~ccl to
soil shari’s of ,~lx’lat-dand in’ igli shin-i,
ran add the Hark pilitits iu oni’ list of poten-
tril pot’tlolio pa\’tilfs.
- exalT) pie, two ~cr tops in a t h ree-di inerl sioria I
l”inalh It we at-C’ also aIiot~ccl to sell fr-ar’-
liotial trllolitlts,,hor-t Vt I’ c-au ~idtl thr- t’niii’tu
gr’e~’~lladed r’egior i Ri ft Lit’ leasih] I’ St ‘ t . e
can a i-ru rlgt! ant’ pat-oft c’otnthii tatiorl ill the
ttvo-dinnerisional plan’’. ltl this case, U ht’ti atit’
sort of addition and st:alat’ iiiultiplit’aticiti is
‘This s also catted ‘‘scatar multiplication,” to distinguish it
trom “vector multiplication.” n whrch two vectors are
multiplied.
tn illttstt’ati’ n’ht and hon complrtetless might
he aluahie to its. let itll~OsO saint’ moore condi—
tiotis on I’oi’ttolio Don us t his lust example ii—
Iustrates buIlt an incomplete st stetil at markets
artd a i’t’dctndaut ht’t \ bet is rd oat/ant it its
pa,t offs ri ft Ot’v otitt’ottte tan hi’ duplicated ht
t
Readers familiar w’th linear algebra should note thai
redundancy is aefi-led as the linear depencenc~of lh~set
o’ payofi vectors Roughly ~eak’ng. if the market s
FEDERAL RESERVE BANK OF St LOUiS
—l I
allowed. nc ~a,t n t’ ran i ise all lunar t art—
h;nanvns ot the ~ LI r’i Lies, ‘I he pavoif sector’
1c’,’ a lineai’ (‘otlll)matiun of tile sN’Llt’i[iCs n ill
aln a~’~ take tilt’ torni: P~ = x~P~ —r
n-bet e ‘a and x, ate real nuiiuht’t’s. Ii LIu’t C!
a i’e no rest rh ti oils on ‘~ a rid ~~. tI ~en all
litien’ i-t)IllhillitlOtl” triO a1lm~pil.
In out- e ~arnp Ic. the pat- ott ect or~ate
chosen so they tin rlot iii’ along the sanir’ line,
‘I~dp~rribc-this nori-pai’alfel quality n e sat
the In 0 ‘~ector’s dVt! linear/v indept’iu/ent. If
111ev did lit’ along We s,ime Iitw, SiiV iflsteEld
that P, = 21’, = il—tat. then tile ~trctoi’s Would
he linr’a,’lv ciepcririent In tin’ east’ of lini’a rh
dt’perlilerlt ef’tors. U C c’ariilot arrange port-
Folios with WI~pa oil cOtllhili?JtiOtl in the
plane’ it’s as it’ \\ C only had a stogie scc’ut’it~
IC) Choose from. \\ hell the t\%o \ t’(’tilF’s ate
lineat’iv indt’pt’tic.hent, so that ant’ payotf c’aii;—
hijanon is aei-t’ssihie, Liii’ tn 0 \ t’t tOrq are
said to span tin’ plane.
~Il of tllis genet’ali~.estliret’tlv to situations
-
‘ ‘
onh’ dltfet’etlcE’ is that it is rr,ort’ dilln’uli to
tri’aph thit’et’ states of the war Id, and tn ipossi-
h
hit’ to “t’aphi totit’ or’ inure. Vt hate~ti’ tile
r’
tlLtllll)et’ of diilieilsiofls, a collection of ‘~ec’toi’s
al~~i•~’s sPatls sonle’thing: tilt’ ClLiCStintl of’ in—
terest for’ us is what that ‘~orl1t’iiurlgis. For
spate (‘all s~ittiat worst a single poitit - tilt’
Oi’i~4iL1 tlc’t I’, = P~ (0,0)1. At most thin spati
a bet o-diinerl SlOt1 al plane. Irl cither ease, ml l~J ‘e
i’erriains an infinity of points ill the three-
diTileilsiorlal space that claritlot lie r’e~tt’heclIn
any c’omhitiations (It Ott’ U’ I) \Pd’tol’s,
hInting or selling ‘.oiiit’ t’oitihittatioti of the
other availahle hi ts the st sietn ut rtlai’Lt’ts is
int-ompli’It’ it the ttuttihi’r iii outcomes i’s—
teeth tile tlutnhc’r’ tit tltln—t’t’dtilltlatlt ht’ts.
later ott ne shall replace one ui these
t’t’dundatlt hr’ts n ith a nun—t i’durldant bet
thus cuttipli’tiog Ill’ market.’
‘rconplete, our payoff voclors are rest.,cted 10 a
‘“arland” within tne larger space of possible statt’ cOrns
S
 39
I
U

I Table 1

BET ODDS PAYMENT T-M-C Tc-M

PAYOFFS BY OUTCOME
M-T-C M-~-T c-T-M c-M-T

I Twins
T places
M wins
7-3
2-3
4-1
—$2
—52
-$2
S667
~ ~
0
‘‘T’
1°f,
$667
~ 53,33
6Too
‘
0
0
‘T”i~T”’o
5333
‘t
0
0
a” —

! 1
I M places
C wins
C places
9-1 1
1-i
3-17
— $2

—$2
—52
$3 64
0
0 :
-
a
0
$235
--J
53,64

0
oj
‘
53,64
0, -
$2.35 -
0
$400
$2.35
[
53,64
$400
$235

I, Suppose that the race is the three-horse affair It turns out we cannot achieve this payoff
described earlier, To make things challenging, vector from the available bets, an upshot of the

I further suppose that the resident bookmaker

will accept only win bets and place bets, refus-
ing to take the seemingly more complicated
fact that this system of markets is not complete.
In practical terms, if the system of markets
were complete, then we could get higher odds

1 quinellas, trifectas, etc.°Finally, to make com-

pleteness an intriguing proposition, suppose
that we get a hot tip from our pal in the stables
on a bet paying off in the event M-T-C than we
can get with the current incomplete set, as will
be demonstrated below,

1t that the race is fixed: the final outcome will be

M-T-C. Because we trust our pal, we now want
to bet only on this single state of the world,
‘rhe key question for market completeness is:
To see more clearly what this means, let’s ex-
periment with the numbers, Suppose that the
bookmaker gives odds on the six allowable bets,

I can we do it, using only win and place bets?

First, note that we aren’t satisfied with a bet
implying the six payoff vectors listed in table 1.
These payoffs have been rounded off to the
on Mastercharger to win: in this bet, we would nearest penny. They include the initial wager of
also be buying a payoff in state M-C-T, which $2 plus any profit from the wager, which
we’re convinced won’t occur, With consistent depends on the odds,”
odds, our payoff per dollar wagered can’t be as

Ii high betting Mastercharger to win (i,e,, betting

on the event, [M-T-C, M-C-T]), or betting Tricky
Bond to place (i,e,, the event, [T-M-C, T-C-M,
M-T-C, C-T-M]), as it would be on the M-T-C tn-
it
ii’ fecta bet,bo We want a portfolio of bets with a
payoff vector that looks like this: (O,O,x,O,O,O),
where x is some positive number, to make the
To say that one of the bets is redundant
means that its payoffs in the six outcomes can
be replicated by an appropriate portfolio of the
other bets,’2 The number of outcomes (six) equals
the number of bets (six), but exceeds the
number of non-redundant bets (five); the system
of markets is thus incomplete. Any payoff vec-

1. maximum profit on our bet, The number x

should equal exactly the payoff we would get
on the M-T-C trifecta bet, if only the bookmaker
tor attainable by combinations of the six bets is
still attainable if we disallow one of the redun-
dant bets and include combinations consisting

I would allow trifecta bets,

9
Show bets are superfluous here, because in a three-horse
only of the other five,

“To convert between odds and payoffs, use the formula:

race, alt show bets automatically pay off. Thus, they’re not p = b(1 +0,), where p is the payoff, b is the size of the
really wagers at all. Also note that, although a trifecta bet (e.g., $2), and 0, is the odds ratio (e.g., 7-3 odds imp-
might seem complex, it is in a sense the simplest bet ly 0, = 7/3 2.33).
here, because it pays off in only one state. This would not “Redundancy requires that we be able to take bets as well
be true of tnifecta bets if there were more than three as place them. By taking bets, we can arrange for a nega-
or more than one race, tive payoff in certain events. For convenience, we assume
10
horses that the bookmaker at Portfolio Downs meets this need by
”Consistent” simply means that the odds on any event
are inversely related to the probability, in the bookmaker’s placing bets at his posted odds. It is also convenient if we
view, of it occurring — the more likely an event, the lower assume that bets can be both taken and placed in any
fractional amount, allowing us to fine-tune our portfolio.
its an
theodds.
event,
odds Consistent
e, byforthe
ratio formula:
theodds
eventaree,0,related
= (1/P(e))
and P(e)to isthethe
—1,
probability
where 0,of is
probability
that e will occur. With consistency, the probabilities of all
the individual outcomes add up to 100 percent,

MARCH/APRIL 1991
S
 40
I—
Table 2
PAYOFFS BY OUTCOME I
BET ODDS PAYMENT T-M-C Tc-M !M-T-C M-c-T :~-T-M C-M-T
Twins
Tplaces
Mwir,s
7-3
2-3
4-1
—$24
$24
—816
580001 580,00
-S40,00!—$40.00
0 0 -
-S~0Oj0
$8000
!~
$8000
0
Lz±~P±0 - 0 -
0 - -

I
M places
Gwins
Total:
9-11
1-1
822
—$40
- $34
-$4000.
0
0
0
0
$40.00
-340,00, —$40,20_i,_,_,~
0
0
- 0
$40.00
- $8000
$40.00
; —640,00
$8000
540.00
I
For example, we can construct the sixth bet tunities, it turns out that, in our example, the
I
as a portfolio of the other five, Consider first a six bets are mutually redundant: any bet that is
$34 bet on Charge Me Interest to place” Ignor-
ing the rounding error, 17($2,35) = $4000, and
our payoffs would be (0,$40,0,$40,$40,$40) un-
omitted can be reconstituted from the remain-
ing five, The same procedure that was just used
to reconstitute the sixth bet could be applied to
I
der the six possible outcomes, Now consider a
portfolio that consists of different amounts of
the other five bets, either taken (negative pay-
generate any of the individual bets from the
other five,” Let’s drop a bet then and replace it
with a non-redundant bet-
I
offs) or placed (positive payoffs), In particular,
For example, suppose the bookmaker does not
let the amounts be those listed in table 2, For
example, we take 12 $2 bets on Tricky Bond to
offer a bet on Mastercharger to place. Instead
of that bet, he gives 3-1 odds on a trifecta bet
‘I
place, while placing 12 $2 bets on the same
on the ranking C-M-T, ‘The odds and the implicit
horse to win, The payoffs to our portfolio are
given by totaling the six columns in the table,
The portfolio yields the same result as if we
payoffs now available to us are given in table 3,
‘rhe fourth row of the original payoff array has
been replaced by the payoffs to the new tn-
I
had placed a $34 bet on Charge Me Interest to
place. In fact, since we placed $80 in bets while
taking $46 worth, our net investment in the
fecta bet, The question is still whether we can
arrange a portfolio of bets that pays off only
when the ranking is M-T-C, That is, can we syn-
I
portfolio is $(—80+46) = —$34, The fact that
equivalent payoff vectors require the same in-
vestment reveals that our bookmaker has set
the odds (i,e,, payoffs) in a consistent way.
thesize an M-T-C trifecta bet? More generally, is
the system of markets complete? The answer to
both questions is yes. The system of markets is
I
complete, which implies that we can achieve
One way to look at the redundancy of one of
our bets is that we cannot synthesize the trifec-
ta bet that we’re after. Even though we would
any payoff vector, including the payoff vector
that corresponds to an M-T-C trifecta bet, I
To synthesize that bet, combine the bets as
place such a bet at any consistent positive odds descr’ibed in table 4. The result is a net two-
(because we’re convinced we know the outcome~ dollar investment that only pays off if the final
that bet is neither offered directly nor can we outcome of the race is M-T-C, The payoff in
synthesize it from the others, In other words, that case is $40, implying 19-to-i odds,” By
we can’t get there from here, creating a bet that pays off under such narrow
Gompleting The Market circumstances, we have maximized the return
on our $2 investment (assuming our pal in the
Another way to look at this is that the book-
maker’ can drop a redundant bet from the list
stables is trustworthy!), With a complete system
of markets, of course, we can also generate a
I
of allowable bets without affecting our oppor- safer portfolio—that is, one that pays off in

“Using 17 two-dollar tickets ($34) simply keeps all the

following payoffs in terms of nice round numbers. We
reconstituted from (1,1). If (0,1) is dropped, however, it
cannot be synthesized from (II) and (2,2).
1
could just as welt scale all the amounts and payoffs down
by a factor of 17 to show the same result,
“This is not necessarily the case with redundant bets. Con-
sider, for example, three redundant bets on two outcomes:
“Again, the scale of the payoffs is unimportant here, Forty
dollar amounts are used to keep the numbers consistent
with the previous example, The important result is that the
portfolio pays off only in one specific state of the world.
I
(0,1), (1,1) and (2,2). If (2,2) is dropped, it can be

I
FEDERAL RESERVE BANK OF St LOUIS
a
 41

Table 3
PAYOFFS BY OUTCOME
BET ODDS PAYMENT T-M-C T-C-M M-T-C M-C-T C-T-M C-M-T

N?S\ ‘4 $2~~~$323N \s\\$fla ~ ‘~NO~NC IN’

\~ \\\t S 0 fl’Oa” •‘I&OG S S

—$2 N” “o
‘~ t4”” ‘St P “ a N’ *ot aoqN~
teN \\~\\12bN ~ $~ $~3~
N N N N N N N ~N
N \NN\\\ N “N ‘‘N’’ N’~
N N N’ N
N N N “, N N N N N
N N’ \N N ~ N
NN\ NIN ~ \NN’N \N\ ~N’~NN’ ~N, ~ ,ç~ ‘~“\‘\ N NNNN NN / N
NNNN’ ““N” N,\ \N\ N N \~‘NNN~

N NNN ‘IN N’’, N

N N’II I~ vN ~“I‘ “ MV4F~$BY~QfltC0ME N

N N ODDS ~PMS$, T~-M~C 74*” fl~ N N e4~M

~ *)~,/t*~N ~SOGNN,/3/~//A,PK2YI,S ~/‘KØ~~~
N N/N eN N ~~‘,fl4> søi~ S4øaN ,$p~ p N’$4oQ~ N

* N~4jNN ~ N NN~ N’flNN S 4

NN”~Mttø~ N”t ~sit p NpN N

/ 34tN$Q 4 ND
o “0
N N ~SflG
0,. ~
tiN
‘I”’

ThtØIN’N N ~ O’N “ ‘S
N N~NN~ NN~\ N NNN N
NN N N N NNN’ ,,\\\N NNN\\ N + N N’

more states—but this would be more costly. For What if there are more unknowns than equa-
example, to get the same payoff in one addi- tions (i.e., more available bets than elements in
tional outcome, M-C-T (i.e., to achieve the payoff the desired payoff vector)?” In that case, at
vector (0,0,$40,$40,0,0)), would require an $8 least some of the bets must be redundant, Thus,
bet on Mastercharger to win, a four-fold in- in determining whether the system of markets
crease from the investment required for the is complete, it is not safe simply to count equa-
trifecta bet, tions and unknowns; we must find the largest
collection of non-redundant bets, In our exam-
It is no accident that the number of bets in
ple, if we can find six non-redundant bets, then
the portfolio equals the number of possible out-
1 every payoff vector is the result of a unique
comes, °Every portfolio we construct is a sys-
combination of these six bets. Thus, in our ex-
tem of six linear equations in some unknowns,
namely the amounts to put into each of the ample, the only way to achieve the payoff (0,0,
$40,0,0,0) is to combine the bets in the amounts
allowable bets, In other words, we start with
($12, —$24,0, —$10,$20,0)
six state-dependent payoffs (the desired payoff
vector), and we seek a combination of weights Finally, there is one last bit of terminology
(ic,, investment amounts) for the available bets which appears frequently in state-preference
that yield those six payoffs. To ensure that such theory. In our example, a trifecta bet has a posi-
a combination exists, we need at least six un~ tive payoff in one state only; in all other states,
knowns (ic., available bets) to work with, Fur- its payoff is zero, Notice that all other bets con-
(N
thermore, some collection of six of those avail- sist of various collections of trifecta bets. Clear-
able bets must have payoff vectors that are lin- ly then, a set of six different trifecta bets would
early independent. produce a complete system of markets, This

I “An allowable bet is included in the count, even if the

amount wagered on that bet is zero, The decision to
wager nothing on a particular allowable bet is an implicit
portfolio decision,
I “‘We shall see below that this situation is not usually a prac-
tical consideration, The normal problem is that there are
not enough unknowns (available bets) rather than too
many.

MARCH/APRiL 1991
a
 42
ii I

leads to the notion of a reference bet or pure oped a simple context in which to introduce the
security. A pure security is a normalized bet
that pays off in only one state. By normalized,
we mean that the payoff in the selected state is
terminology of complete markets. An analogy
can be made between Portfolio Downs and real
markets. Bookmakers manage their risk by lay-
I
ing off bets, while investors manage their risk
one unit)8 To get a normalized payoff) we must
adjust the amount invested in that bet. The
wager amount that implies the normalized
by hedging their portfolios; bettors are unsure
whether Tricky Bond will win the race, while
I
payoff is called the price of the pure securicv. investors are unsure if pet rocks will be
Any state-contingent claim can be regarded as a
collection of pure securities. A system of mar-
kets is complete if and only if the number of
popular with consumers; etc.2°Our next task is
to flesh out this analogy to see what importance
state-preference theory can have for more gen-
‘II
attainable pure securities (either directly or
through combination of other securities) equals
the number of outcomes.
eral economic analysis. This is done with a
series of examples. 1I
Multiple Periods~Multiple
CommoditIes
SOME ECONOMIC APPLICATIONS
One of the primary insights of state-preference
Historically, the theoretical economics
literature has generally followed two distinct,
but entirely compatible hnes in interpreting and
theory is that the traditional notion of what
constitutes an economic good can be readily ex-
tended in a way that allows us to examine
II
applying the theory of complete markets. First,
led by the initiators of the theory) there were
applications to the problem of general equilib-
economic behavior across time and under un-
certainty. At the same time, this extension
forces us to think of ‘goods” in a very different
ii
rium. Most work in this area now starts with way. Our notion of a good is broadened in two
the notion that markets are not complete and directions: time and state.
proceeds to analyze the nature of equilibrium
Examining the time dimension first, consider a
(or disequihbrium) with incomplete marketsJ~
simple example involving a banana and an apple.
The other line of research has focused on the A textbook would tell us that each consumer
relative efficiency of financial markets at alloca- has a set of preferences such that she either
ting risk by providing greater investment flex- prefers one to the other or is indifferent be-
ibility to investors, in the same way that a com-
plete system of markets makes bettors better
tween them. In our new way of looking at
things, however, we must include a time dimen- Ii
off at Portfolio Downs. In practice, the general sion in a description of the commodity. From
equilibrium applications tend to de-emphasize
uncertainty and concentrate on production, con-
sumption and intertemporal optimization. Con-
this new perspective, “apple” is not sufficient to
describe a good; one must specify either “apple
today” or “apple tomorrow.” Merely specifying
ir
versely, the financial market applications tend “apple” is analogous to stating “red” without
to ignore real resource constraints and temporal
factors, in order to concentrate on uncertainty.
specifying “red tricycle” or “red Ferrari.”
To make the example more specific, suppose
F
The first section of this paper considered the that our consumer generally prefers apples at
properties of betting odds at an imaginary race- time t, A,, to bananas, B,, but that variety is also
track. Our goal in that example was to find valuable to her. In this case, the following pre-
some conditions that would ensure complete in- ferences for consumption over two consecutive
vestment flexibility. In the process, we devel- days might hold:
IBAt Portfolio Downs everything is measured in dollars, so a For example, a bookmaker who takes a large bet on some
pure security would be a trifecta bet that paid $1 if the event from a bettor can lay it off by placing a bet on the
outcome was the ranking specified in the bet. Pure secur- same event with another bookmaker,
ities are also called Arrow-Debreu securities or primitive
securitie&
19
See Geanakoplos (1990) for some examples.
20
The terms “laying off bets,” ‘covering a position,” and
‘hedging or reinsuring a portfolio” all refer to the same
basic process of reversing a transaction with one party
I
by making a countermanding transaction with a third party.

1
FEDERAL RESERVE BANK OF St LOWS
a
 43
F
Ranking Bundle Sequence of fruits apple and banana tomorrow; we shall eat only
I 1
2
(1,0,0,1) apple today, banana tomorrow
(01,1,0) banana today, apple tomorrow
one or the other,
Because the additional consideration of time
3 (1,1,0,0) apple today, apple tomorrow and uncertainty forces such a radical shift in
I 4 (0,0,1,1) banana today, banana tomorrow,
where the bundles (i.e., payoff vectors) are of
our conception of commodities, it is worthwhile
to consider it further. The distinctions between
apples and bananas, today and tomorrow, and
the form: (A0,A,,B0,B). With our new definition
I of commodities in mind, the four sequences
ranked above can be treated under standard
peace and war serve as a simple basis for an
example Suppose that, today, the political situa-
tion is peaceful, but tomorrow’s situation is un-
utility theory exactly as four commodity bun-
certain. This implies two possible states of the
I dles, each composed of a pair of the four dif-
ferent commodities. One might consider, for
example, the indifference curves between ap-
world: peace today, peace tomorrow, and peace
today, war tomorrow. We thus have six com-
modities, abbreviated as follows:23

I ples today and bananas today, or budgeting

between apples today and apples tomorrow.
An infinite number of other bundles could
Commodity
apple (peace) today
Abbreviation
A—•--0

I also be described and ranked, of course. For

example, (15,0,3,7) is a possible bundle, one that
would be preferable to any of those listed
banana (peace) today
apple peace tomorrow
apple war tomorrow
A-P-I
A W -1
-

above. Restricting the time dimension to two banana peace tomorrow B—P—i
banana war tomorrow B W —1
—
days and considering the resulting four goods,
we see that the system of markets defined by
Suppose now that the local wholesaler, Whim-
the four bundles here is incomplete: we can get

IL bundle 1 by buying bundles 3 and 4 while sell-

ing bundle 2. Thus, there are at most three
non-redundant markets for four goods.2~
sical Fruits, sells the “fruit baskets” or state
claims described in table 5.
This admittedly contrived example (note, for

II Uncertainty
The same approach is used to incorporate
uncertainty. To do so, the definition of a com-
modity is expanded to include the state of the
world. Thus, for example, an umbrella in the
example, that basket No. 4 consists of 364 ap-
ples today and 364 apples tomorrow, plus 364
“bonus” bananas to be delivered only in case of
war tomorrow) allows for ready interpretation,
because of its similarity to an earlier example.
The array of goods here is essentially identical
to the payoff array for Portfolio Downs; only

Ii rain is now a fundamentally different commod-

ity from an umbrella on a sunny day
22
Simi-
larly, “bananas in peacetime” are different from
the labels and the scale have been changed. At
Portfolio Downs, the only distinction between
goods was the state of the world; the physical
commodity in all cases was cash, and time dif-
“bananas ferences did not exist. At Whimsical Fruits, on
however, in wartime.” This
fundamentally new application,
expands our notion of
a good, Because states of the world, by defini- the other hand, the goods have different inter-
pretations, and all the amounts are scaled up by
a factor of 100; otherwise, the payoff array is
tion,
the are mutually
notion
of physical of exclusive,
economic
consumption. For we mustfrom
consumption
example, separate
that
if we identical to that in table 1, We can therefore
purchase the bundle of goods consisting of “an conclude immediately that the fruit market here
is incomplete, For example, a buyer wanting on-
apple in peacetime
wartime tomorrow,”tomorrow” andbe
we shall not “a eating
banana in
both ly to buy apples to be delivered tomorrow in

f*1 21
There is, of course, no reason to limit the time dimension
to two days. By extending the time dimension indefinitely,
we get an infinite number of goods, which would require
an infinite number of markets for a complete system.
22
commodities dated today, the true state of the world is
uncertain, although the current political situation is known
to be peace. In principle, therefore, we should distinguish
between apple-peace tomorrow-today and apple-war
tomorrow-today. As a practical matter, however, we cannot

I
street vendors in New York, for example, charge $3 for an observe this distinction until it is too late for it to affect our
umbrella under clear skies, and $5 for an umbrella when
behavior, For convenience, the distinction is not made in
it’s raining, the example.
23
Note that there are only two states of the world, distin-
guished by the political situation tomorrow. Thus, for the

I
MARCH/APRtL 1991
a
 44
I
Table 5
NUMBER OF FRUIT IN EACH BASKET
I
BASKET PAYMENT } A-.-O 8-.-O , A-P•1 A-Wi B-P-i B-W-i

2
3
1 —$200
—8200
—$200
L,_667
333
0
.667

i
333
0
, 0
333
1000
0
0
1000
0
333
0
i
I
0
0
0
I
4
5
8
—$200
—$200
—$200
364
0
0
i
~‘
0
0
235
÷
‘ 364
0
0 ‘
364
0
235 .
0
--——

400
235
364
400
235
-

case of peace cannot arrange such a transaction
as a mixture of the baskets offered at Whim-
sical Fruits.
Do Complete Markets Really Exist?
We now have a sufficient context to ask
whether markets are complete in reality. The
answer from economists who have considered
this question is a resounding “No.”24 The prob-
lem is the huge number of markets that would
be required for completeness to hold, Even with
the roughest distinctions (e.g., measuring time
in one-year intervals, considering all automo-
biles in a given year as perfect substitutes, etc.),
an astonishing number of states of the world
must still be considered. For example, every
conceivable future invention must be included.
Furthermore, the timing of invention is signifi-
cant; if a cure for cancer comes in 1.997 instead
of 1998, that implies a different state of the
world. Some would have us distinguish goods
by geographical location and even by the iden-
tity of the final consumer! 25 We need to mul-
tiply the number of states by the number of
periods and then by the number of physical
goods and services. Finally, we need a market
with which to exchange every one of these
many goods with every other one.
The absence of many markets from a real
economy may be explained by transaction costs.
For example, a contract for “an apple dated
I
2006 if a cure for cancer is discovered in 1997”
I
is too costly to arrange, relative to the marginal
benefit of such a transaction. Such a commodity
is too narrowly defined to be of interest.
I
Turning to a more plausible example, the
International Monetary Market (1MM) of the
Chicago Mercantile Exchange (CME) established
~1
an organized marketplace for several commod-
ity futures contracts in 1973. Among these were
contracts for (1) bagged silver Canadian coins
I
worth C$ 5,000 at face value, and (2) 12,500,000
Japanese yen.2°While futures trading in silver
coins was discontinued several months later
JI
after trading volume dwindled, yen futures still
trade today with thousands of contracts chang-
ing hands every business day. ii
In hindsight, it seems clear that the conve-
nience provided by a market for bagged Cana-
dian silver coins was outweighed by transaction
1I
costs of some form.27 On the other hand, the
successful introduction of a market for yen fu-
tures suggests that investors would have faced
significant portfolio constraints in their absence.
11
The contrast between these two raises an inter-
esting question: are the incomplete systems of
markets observed in the real world efficient in
II
the sense that the missing markets are absent
because they are not operationally cost-effective
given current trading technology? Or, are they
incomplete simply because no one has thought

24
fladner (1982), p. 930, for example, states that”... it clear-
ly requires that the economic agents possess capabilities
of imagination and calculation that exceed reality by many
orders of magnitude.” Geanakoplos (1990), p. 2, states
that, “There is little doubt that permitting the incomplete-
ness of asset markets is a step in the direction of
realism.”
25
See, for example, Geanakoplos (1987), p. 116. This last
distinction is made in analyzing public goods and
externalities.
26
5ee Chicago Mercantile Exchange (1974, 1989). All futures
contracts specify a future time and place for delivery, in
addition to other standardizations. The state-contingent
nature of markets for futures and options is considered in
greater detail in later examples.
27
0ne possible explanation is that the payoffs to coin futures
lie in the space spanned by linear combinations of Cana-
dian dollar futures and silver futures, making the coin
futures redundant.

FEDERAL RESERVE BANK OF St LOUIS

a
 45
Ii
U
to provide the services required to match buy-

I ers and sellers?

A full answer, unfortunately, is beyond the
Figure I
The Allocation of Pork
scope of this paper, although we return briefly

I
FAT (JACK) JACK
to the issue below when considering futures
markets. It is not a simple task to identify be-
forehand which missing markets impose the

I most significant constraints by their absence,

and thus would be most likely to succeed with
investors. The theory of complete markets does
tell us, however, that, contingent on the level of
I transaction costs, identifying and providing such
markets can make everyone involved better off.
LEAN
(BUBBLES)
LEAN
(JACK)

Not surprisingly then, much of the theoretical

I work in this area investigates the properties of

economies where markets are incomplete. Some
of the issues involved are presented in the fol-
lowing examples.
I 2 3
complete Markets and Efficient BUBBLES FAT (BUBBLES)

Allocations
I We first examine one of the most basic pro-
blems in economics: how to arrange for the
and Bubbles gets all three fat portions.3° Despite
best allocation of available their differences, Jack and Bubbles recognize
the nursery rhyme of Jack resources.
Sprat, but Consider
suppose
that Jack and his wife, Bubbles, are now di- the value of trade. For example, the vector \~in
vorced and living separately.25 For simplicity, the figure represents a sale of one pork chop
from Bubbles to Jack, where Jack owns one
let’s ignore
The uncertainty
local market and theon
has a special time dimension.
pork chops, pork chop before the sale and two pork chops
which consist of one fat portion and one lean afterward.
for $P per pork chop.29 This can be Because there are two commodities but only
portion,
represented as a single payoff vector: one transaction (i.e., one payoff vector) in this
example, the set of markets is incomplete. The
COMMODITY
II PRICE FAT LEAN
single available market spans the 45-degree line
in the payoff space of both consumers; each
consumer can achieve any payoff where the
L1
number of fat and lean portions is given by
I Now consider the allocation of a fixed set of
pork chops, say three chops. between Jack and
some multiple of (1,1). Unfortunately, the opti-
mal point, A”, does not lie in the space spanned
Bubbles. We represent this situation by the by this available market. The upshot is that the

.1 modified Edgeworth box in figure 1. Given the

special preferences of Jack and Bubbles, there is
a unique optimal allocation in this example, rep-
system of markets is incomplete, implying a sub-
optimal allocation of resources. In other words,
the allocation of the six (fat and lean) portions

I resented by the point A” in the lower-right-hand

corner of the box. This is the point at which
Jack gets all three lean portions of pork,
here ideally would be represented by A~the
unique Pareto-optimal allocation. The available
market does not allow them to trade to this

I 29
Recall that: “Jack Sprat could eat no fat I His wife could
eat no lean I And so betwixt them both I They licked the
platter clean.”
be made better off without harming someone else, In this
particular example, there is only one Pareto-efficient
allocation, because of the extreme nature of the prefer-
29

I Note:”To market,to market,to buy a fat pig/Home again,

home again, jiggety-jig.”
30
The optimal allocation is called Pareto-efficient by
ences. In a less extreme case, there would be many
Pareto-efficient allocations, depending on preferences, in-
itial endowments and relative prices.

I
economists. With a Pareto-efficient allocation, no one can

MARCH/APRIL 1991
a
 46
allocation, however, and Jack’s fat goes to
waste, as does Bubbles’ lean, Now, let’s complete
the system of markets by adding a second al-
lowable transaction: say that Jack and Bubbles
agree to trim the fat from the chops and sell it
separately to one another at a price of $P/2 per
portion of fat, Thus the set of payoff vectors
becomes:
COMMODITY
PRICE FAT LEAN
1 1
$(P12) [1 0 cents per pound ($35.50 for a 50-lb. bale), and
The vector \‘ in the figure represents the sale
of a single fat portion from Jack back to Bub-
bles. By thus completing the market, we have
made the optimal allocation attainable.”
This same principle of allocational efficiency
can be seen at work in more general settings.
The major contribution of Arrow and Debreu
was to show that an efficient allocation of all
commodities is feasible for an economy with
complete markets, even with many periods,
many physical goods and services, and uncer-
tainty about the future.” We next illustrate the
importance of complete markets in this context,
without considering a full general equilibrium
model.
Future8 Markets and Risk-Shjfting
Uncertainty is a salient concern in many mar-
kets, It is common for contracts to require
alternative payoffs in different states, Commod-
ity futures contracts, for example, specify the
exact physical characteristics of the commodity,
the date of delivery and the location of delivery;
moreover, they typically provide that ‘If deliv-
ery or acceptance or any precondition or re-
quirement of either is prevented by a strike,
fire, accident, action of government or act of
God,” the directors of the exchange will decide
the duties of buyer and seller.”
From a practical perspective, ho\vever, the im-
plicit state-dependent nature of such contracts
is much more important than such explicit stip-
31starting
from the same initial allocation, where Jack has
one chop and Bubbles two, Jack buys both chops from
Bubbles and then sells back to her the three fat portions.
32This is their proof of the existence of an efficient general
equilibrium in an economy with complete markets, See
Geanakoplos (1987) for an overview,
FEDERAL RESERVE SANK OF Si, LOUt S
ulations, Commodity futures contracts enable
owners of a physical commodity to hedge the
value of their inventory exactly against uncer-
tain future fluctuations in price. The state-
preference approach can provide useful insights
into the nature of such markets,
Futures markets allow people to contract to-
day for future delivery of a specific commodity
at a specific price. To see why such a contract
is valuable, consider a hypothetical cotton
market without a futures market, In particular,
suppose that the current price of cotton is 71
there are two states of the world, In one state,
the price of cotton will increase to 80 cents per
pound; in the other, it will fall to 70 cents per
pound. Finally, assume borrowing and lending
are possible at an interest rate of 5 percent
over the period. Using the tools of state-
preference theory, we can set up the payoff ar-
ray in table 6. There will ultimately only be two
cash markets for cotton here. We can either
transact now in the spot market, or we can
transact later in whichever of the two subse-
quent spot markets is available,
The Hedger
Now consider the situation of a cotton farmer
who will harvest 25 tons of cotton in the com-
ing period. To restrict the number of contingen-
cies, we treat the size of the crop as certain, In
terms of state-claims, this is the endowment
(not a transaction) described in table 7. Finally,
assume the farmer is risk-averse and wants his
cash receipts to be the same, regardless of the
state of the world, In other words, he wants
a final consumption bundle, after harvesting
and selling the crop, as described in table 8,
Note that we must restrict the payoff $X to be
strictly greater than $35,000 here, Otherwise,
the farmer could achieve a certain payoff of
$35,000 by giving away $5,000 in the high-price
state of the world, The potential for arbitrage
implies that he should be able to do better than
this, Can he achieve his desired payoff with the
current set of markets?
“See Chicago Mercantile Exchange (1983), p. 8.
 47
III
I

11
II
II
Ii
I
I
I
I
I
I
I
I
I into
Thea sale
ward answer
certain dollar
is no:ofpayoff,
contract tothe
convert
he requires
form his certain
describeda in
for-
crop
sactions. In this example, no price is charged
to store or lease cotton, although such a price
could readily be included, In our simplified two-
state, two-period example, this is sufficient to
I table
can 9.beSuch
able itcontracts.a In
contract
synthesized does
as aof
terms ournot exist discus-
combination
earlier here, nor
of avail-
complete the system of markets. The payoff
array is now given in table 10. To arrange the
sion, the system of markets is incomplete. In desired bundle, the farmer can now arrange the

I this case, the incompleteness prevents the

farmer from arranging his desired payoff vec-
tor, Let’s consider two ways of alleviating this
constraint by completing the system of markets.
First, let’s add storage (and leasing) of the
physical commodity to the list of allowable tran-
transactions in table 11, converting his endow-
ment into a certain dollar amount next period.
In other words, the farmer leases 25 tons of
cotton, sells it for cash today and invests the
proceeds, repaying the borrowed cotton when
his own crop is harvested. Thus, it is possible,
even in the absence of a futures market, to hedge

MARCH/APRIL 1991
a
 48 ‘I
Table 10
STATE CLAIM PAYOFFS
TRANSACTION CURRENT Cotton Cotton Cotton Cash ‘ Cash
COST Now Later Later Later , Later

Buy now
Buy i~ 70’
.535,500
0
th 706

25 tons
ci 80’ ‘‘1 70’

0
4 --$35000
‘~eo~

0
-

Buyc~B0’ 0 0 0 25t0n5 9.~

Borrow cash $100 0 .t—-.
0 ,
t..-
0 --- •i
—SW5
..-
—$105
Store cotton 0 —25 tons 25 tons 25 tons 0 0

Table 11
STATE CLAIM PAYOFFS
TRANSACTION CURRENT Cotton Cotton Cotton Cash ‘ Cash
COST Now Later Later Later Later
~ 7 c 80’ ‘~ ‘U 70’ 2z so:
Lease cotton 0 - 25 tons’
0
—2stons 4 ._T
25
t
2~~
Sell now $35,500 —25 tons ~ 0: 0 40 I -. 0
Lend —535.500 0 0 0 $37,275 ‘ $37,275
Total: 0 0 — 25 tons —25 tons $37,275 537,215

Table 12
STATE CLAIM PAYOFFS
TRANSACTION CURRENT Cotton Cotton Cotton Cash Cash
COST Now Later Later Later Later
w 70 (ci 80’ @ 7Q6 cit 80’

Ritures 0 0 —SOIb. J —501b

the crop. ‘his is achieved, however, through a
circuitous and potentially costly chain of three
transactions. Why shouldn’t the farmer arrange
the desired transaction directly as a single
contract?
- than one way to complete an incomplete system
This is precisely the role of a futures contract,
our second means of completing this system of
-‘
markets. A futures contract would have just
thts payoff vector, perhaps adjusted by a scale
- ture that all missing markets result from trans-
factor. A cotton future is a standardized for-
ward sale contract; i.e., a contract to pay a spe-
- - . -
cified price (P) for a standardized quantity (oO
- - . ever, is beyond the scope of this paper.
lbs. of cotton) at a specific time (next period), re-
gardless of the state of the world, as described
in table 12. The futures contract makes the
marketplace more flexible; in our simple exam-
pie, it completes the system of markets. It
allows the farmer to transfer directly the price
risk associated with commodity ownership with-
out transferring ownership of the commodity
per se.
There are three lessons here. First and fore-
I
most, completing the system of markets makes
everyone better off (or at least not worse off)
by allowing risk to be transferred from the
farmer to a speculator. Second, there is more
, -
of markets. rhird, some means of completing a
-
. system of markets may be more cost-effective
. - -
than others. One might even plausibly conjec-
-
-
action costs that render them cost-ineffective.
Confirming or refuting such a conjecture, how-
-
The Speculator
The same transaction can also be considered
from the perspective of the speculator who
buys a futures contract from the farmer. A
speculator is someone who wagers that she can

FEDERAL RESERVE BANK OF St LOUIS

a
 49
II
I

N’ <>7 N ~ / N ‘N ~i\N\ NNN N \NN\ N/ NNN 7NNN ~N N

I Tab~t

,m*tiSt”
7
>7<>
5
‘N’
/
s’7,~
/
<>N< 7/~7*\<N<NN/~
7/

>11~S~’
>>, ~/
~tATECLMMPA~PS<
f” Casfr I
~“
~~~<>:~‘
N
~N/” NNN

“7
~cw~RENt
j
C*,tS
7
T4o~t7 LIN~ Ca
L*tá
c~sS>“>I’~,w CasW tN

“V ‘1
11 N <77>7

7/’ 7 /
/
/ ~
71 / ~k
N’N’’’NN
~
, , N~ 5/’’~ N

~, :Y~~:7,7. 7
;~7N’/ 11 9~:t>”< 7 :7 ‘7<4<’ ~‘ ~t0V~’f7 / / LIE
I 7/:
NN(
at~ 74*“N,~/
~

,,N,/>,ç4,N/
~ _____

~/N/I’~/N~7r<//*’
7<~>N<
~

7////5*N
~

:,‘NNtft/L7
71
\~771N
/
iLl
“7’ N 77 “7

I /N7 ,/

iWt4-
7 ~~“7

~ ~
7’

t”7 ~,/<
—~

~
;/N7

~ ~*/7 ~
/NN’7NN

7~S/4,1
‘N/’K\,/NN”, < ‘>~ ‘‘,‘>“ ~
e~4~r~<
“-At~±I”
t~
7

~ ~tca —1 ~ ~ :~-cj
1.,tt1J 7//a”>
I
,N~
‘/7 “
/
‘ ~,
/
t /a
r~pa~ <‘
77
‘~‘‘N*//~t
-~,-~>N c~

“~:~ $ <j: ~
N ,> ‘>~ ~ ~
~

_4~.
/

7’- ~ t-’
~4~-~- < 17
Pa”’
I ‘N , , N*NN7

7’

7”’
7
‘7 ,

4P.SMrt7tXr5*LI>iiL’>~L
N ‘ ‘,/N ~N N N
-
‘>,Nt” NN

----/~
YNNN N / >‘~“‘~

,-~t/ ’,_’
7
~‘ N
“~—,f,-
~t”

fP_~11~
~

I accurately predict the state of the world.34 To This concentrates the speculator’s return on the
see how speculation occurs in the absence of single state-claim, cash later in state P~.

I futures markets, we start with a market that

allows for storage (and leasing) of cotton. If we
omit cotton storage, as in table 6, then the sys
In other words, to speculate on the low-price
outcome in this world, the speculator must sell
the physical commodity short. This requires
• tem of markets is incomplete; the speculator is
leasing the commodity, selling it in the spot
~ thus prevented from arranging her desired pay-
market and repurchasing the cotton later when
off vector. Once again we complete the market
the price has changed. These four transactions
in two different ways: with commodity market

I speculation and with futures market

speculation.
imply a profit if the cotton price falls and a
loss if it rises. Finally, the speculator invests
an amount equal to the discounted value of the
cotton in the high price state. If this state oc-

I To make the problem more general, we now

represent contract quantities (measured in lbs.
of cotton) by C, and prices (measured as cents
curs, then the cost of repurchasing the cotton
will be exactly covered by the investment; if the
per lb. of cotton) by P. There are three prices, low price state occurs, the investment more

I the current spot price P0, low price and a

future high price P (where P >Pj. There is a
11 11
~L
than covers the repurchase of the cotton, and
the difference is a speculative profit. The cur-
rent cost of this basket is the proceeds of the
single fixed interest rate R, at which the spot sale, P C, less the amount that must be in-
I speculator can borrow and lend money. This
payoff array is given in table 13. Let’s say the
speculator believes the low-price state will oc-
11
0
vested, P C/(1 +fl). This must be negative—i.e., a
positive cost to the speculator—if an arbitrage
opportunity is to be avoided.

I cur. In particular, she wants to arrange a trans-

action that pays only cash in the low price
state. She can accomplish this by arranging the
One must question the practicality of such a
transaction, however. Once again, there is more
bundle of transactions described in table 14. than one way to complete a system of markets.

I 34Note that the speculator is not necessarily a consumer of

cotton. Indeed, a cotton consumer (e. g., a clothing manu-
etfect, a speculator sells insurance (i. e., bears risk) for a
living.
facturer) is likely to be as risk-averse as the farmer. In
I
MARCH/APRIL 1991
a
 50
Table 15
TRANSACTION CURRENT Cotton Cotton
COST Now Later
:~PL
Sell Standard Fjtu’es 0 0 —50 lb
Bu~ ~L
‘~ 0 0 50 lb
Buv~P~ 0 0
cash Settlement Fut-jres 0 0 ‘ 0
Although shoit sales are commonplace in the
stock market, for example, such transactions
can be considerably more difficult when dealing
with physical commodities. If a market for cot-
ton futures is available, however, the speculator
can arrange her desired bundle without evei
having to store or lease a physical commodity.
Indeed, most futures contracts are retired by a
process called cash settlement, which obviates
any transfer of the physical commodity.
In effect, a cash settlement futures contract
(table 15) is a bundle of transactions sold as a
single unit, where P. is the futures price. Note
that, to eliminate arbitrage here, it must be the
case that ~F~L > 0 > P P .
6 11
Such a standar-
dized contract facilitates the transfer’ of i-isk
from hedgers to speculators. Moreover, with
cash settlement, the speculator never has to
handle the physical commodity sold in the fu-
tures contract, thus reducing transaction costs.
Options and .tnvestor Fleyibility
‘The preceding example illustrated how the
introduction of a futures contract, a paper
transaction, could improve the allocation of re-
6
sources in an economy.’ Such applications of
state-preference theory are not limited to the
“In general, in order to eliminate arbitrage opportunities,
every payoff vector must involve a trade-off, in which some
element (including the current cost) of every vector is
negative and some element of every vector is positive. In
other words, every vector must represent an exchange of
36
some sort rather than a unilateral gift.
In one sense, we have reallocated risk rather than
resources. Recall, however, that we have redefined goods,
so that cotton in the high-price state is a different resource
from cotton in the low-price state.
“For more thorough analyses of options and complete mar-
kets, see Ross (1976), and Arditti and John (1980).
8
‘ The ‘commodity” in most options markets is really a
share of a corporation’s common stock. However, there
are also organized options exchanges for cattle, copper,
crude oil, Canadian dollars, etc. See the shaded insert on
pp. 51-52 for a basic description of option contracts.
FEDERAL RESERVE BANK OF St LOUIS
I
U
STATE CLAIM PAYOFFS
Cotton Cash Cash
1
Later Later Later
I
‘
@PL
--50 lb ~F ‘
0 Pj.. 0
SOIb , ..•
0
futures markets. Art-ow (1964) showed that the
ability to reallocate risk without otherwise con-
straining economic activity is a general property
of securities markets. This principle can also he
seen at work in the options market.” ln con-
sidering an options example, we abstract from
the issues of timing and consumption and
concentrate on uncem’tainty, to streamline the
example.
Options markets are especially useful for- shift-
ing risks, because of a special characteristic of
an option contract. In particular, a call option
specifies a strike price, labeled K, at which the
holder of the option can purchase the underly-
ing commodity.’~By simply changing the strike
price in an option contract, we can create a
U
fundamentally different financial security. Thus,
a ‘single” options market provides the oppor-tu-
nitv to exchange a multitude of state claims. To
see how this works, consider the following
1
example.
The Chicago Board of Trade Options Exchange
(CBOE) trades options on the Standard and I
Poors’ 500 stock index (S&,P 500), among other
things.” A call option on the S&P 500 with
striking price of 295, theoretically gives the op- I
“The S&P 500 is a weighted average of 500 common
stocks. The amount of each stock included in the index is
pro-rated according to the value of that stock, where the
value is defined as the price per share times the number
I
of shares outstanding. The value of the index is then scal-
ed down to make the base period index (1941 -43) worth
ID units; if the S&P 500 today is worth 295, for example, it
is worth 29.5 times as much as it was in 1943. Thus, the
I
price of the S&P 500 is not, strictly speaking, a dollar
value for the index,
I
a
II 51

II NC~N\P\!\\NN/NNN IN . : : NN/ N N N N: NN NN~N::N~N:

: 4ai~<NNIP~ ~ tth1~èst NN NN ‘ N NNN’ N N NN

W
kit :: :~~:~ g~ git t

N\
bu

?
a

tl>~Mt0
dat
peotfed
‘N ~jfl~

N
a
,
p~
valpeN o~
cati opuqn t>o7Ho1d~r
N’

I . ~ N N, ~byoN\ N
paflyt S epflatttr ,, N

~ /_ 1*7 tad ~i N

W! ~ ~/ tit”
NN7
N/N
>,,N~
‘t 7th ~NN //k>N//b/’N
/ N
7Y( ///N N

op~rfbnp <op at ‘bfrJct, / N N 7 N

I~N” r*7’ “Ø~f/

I 7’ / / ‘~ N’ N / ‘‘//‘~/N7”
N >,/P”N/
/
NN N N N
N N N

‘7, NC / //N <7’7N<>N

5
N N

I 7/1b1* / êI 4 7 / N

N N N

I N///N,N/N/ ~~/N ~<< N/~ NNN N NN

epzn /
~
N %~/ I “N ~, 7 7
7 / N’N/ / NN N’> NN’

N27t b~ / 7>>’ <N 7 “/7

N N,NN/
“N,

5 / “‘‘ 7,, ,, / 7/ N
/ <~‘ N, ,, / NN ‘P~P’~~:’1”N N’ / N ~ 7 “ 7> N”

I $1 b’ / N, ““ NN/>N,NN / N /
7
N, i”
7/7
/
/
‘7/, //N/’~’>/*/
A
v/N/_’//NN ,, N///N/’>NN
// ‘N///’”
N ,N’7’//oNN*7
/
7
N
7”
/ N / // //
/ ‘ 7,’
,7N/,/N_
7, 7’N
‘N /
/
N 7/ N /
‘ / N

/71 ~,/N’ ~ N ~‘“‘

‘>/‘<~ ‘> /7~77>N~~ / / / >1W7.
/ ,N’NN ,/~ /,,pt,,,,,
N/NNNN,/
7
. ,77 V

I /775
7fl~
0’~,,_
/ —<
,~ 7 ‘>7 N,~*N/N/P7/,N
7
i
7 7/
1 /
;‘NN”’’/;
/ N’ / ~N N/N
—

/7>
N
/‘‘N
/
,N
/
NVNNN/ N,’ N N ~ 7/7’,>”
/
/>N~7NN/
/ N’
7,,,,— /~ N,
,N2 /
N’ ‘7 ,,7, ,NN *// /
7//N>
7 7
N 77/ 77 >~ ~
N//N,N/,
/
N///7//N

‘N
nib ,>/~

‘N / 7
,~

N>
N,’>/NN/;NN,/~N

/
/77
~,
j~NN /
,,
,‘/>N7
/
7”
/
; N
5”’>
N

I
• / // , N N N, ~< 7 NN’
N N>> ,/
/
N’ ~N / 7 /
~“N77N’NN
, / N N
/ — “7/,
‘7
/
N
/ // N
N//~//N/N/NN>N/
‘
N’

/ N7,~” 7/7,> “<7’ / N/N, N 7’, N’ N ,/‘>< 77’~/

/ 77
,// NN/7/N
~‘NNN,N ‘N /
NNN7/
4e~.N,’N >//N
NN’ // ‘/NN’71//4
/7 / N N> /N/ / /
NN7’/7/’7’7’7’7’
/ N ~ N7
1’N//*N7:/”~/
‘7/
1N

I S
N”Nl0~S0>~
/ N

N, N,
**1
N N
/$t
‘>thbq
“N
N’

/
I>”~4,,
/
/ a
N, N, /
Nt >~
AqttnC
NN~NN

“
/
‘N
7’
N’
/

/
N”

i <N> tlN’c?~ N \‘~ 71

I ‘N
pa
7
V
txeaf optsVsa%
‘Th 1
IC / N,N’ N
N N>
LI 7”

N/N NN, /NN>NNN7NN N N~ NN NN ‘ / j,>, <N N N N/ N N NNN NN N\’~ N,NN

I
N N N \N N N, ‘>N N/<N N N 75 NNN\N/N, N < N

“tnb <7
‘N N 1 01 ib N N N’,, 7’N NN N

7N’*N tebet~h U~w N f3N~~jfl DpI) ~~~

I ~~<‘!“
<N,
Nftt
ottad
~

gjN
~
Wze~t
‘ttgthesmk
,,h~
edNN a
thlt4osIyNnght
h~ an
Io’Itl
Pt
1WU>U1~~NNN

N~kN N>
NDINN
NNtha~PN1N’~~N

thp
N \\
NN

te’pwana
\ldeN
fltl1
tw
N ~hètb<t’7<
rio
N

Nt~ AU oder
NIhNNOPO NP1
p ~withle
c> ~~M~~NNNN1os dNt1~N~
~ N f tiM N ~p ~j~tN~a N NrNONNVItNN 1’
N

N N P NUOTh~~ N N N: N N hot CNflth~ ;bpgN ~~~j(NN~N<fr

a MARCH/APRfL 1991
 52
‘II
receives in cash $500 times the difference be-
tween the value of the index and the striking
price, if this difference is positive, and nothing
otherwise.”
To simplify our example, we limit the number
of possible outcomes as before and ignore the
time dimension. As at Portfolio Downs, there
are only two relevant times: before the true
outcome is revealed and afterward. Each state
of the world corresponds to a different value
for the S&P 500:

State Value of S&P 500

A 286
B 294
C 299
D 301
E 306
F 314

Thus, our option to buy at 295 is a state claim.

tion holder the right to purchase the S&P 500
index at 295.”°Because the weighting scheme
involved in constructing the index involves 500
stocks, many held in fractional quantities, all of
~‘hich must then be scaled down to a relative
value, however, it is impossible to purchase the
exact S&P 500 index. Instead, cash settlement is
used: at maturity, the holder of a call option
“°The units for the option contract must be rescaled to give
a dollar value. In particular, option contracts are for $500
times the level of the index. For example, to exercise a
call option at strike price 295 would cost $500 295
x
41For example, an investor exercising a call option at strike
$147,500.
FEDERAL RESERVE BANK OF St LOUIS
In the event that the value of the S&P 500 itself
is less than 295, then the option is worthless,
because the (approximate) index can be pur-
chased in the open market for less than the
strike price. If the index is worth more than
295, then the option is worth $500 times the
I
difference between the actual price and the
strike price. This is summarized in table 16.
It should be clear from our earlier examples
II
that this system of markets is not complete.
There is no way, for example, to arrange a
portfolio that pays off exactly $50,000 if the
IF
S&,P 500 is below 300 and zero otherwise; i.e.,
($50000,$50000,$50000,0,0,0). The special char-
acteristic of options markets, however, is that IF
linearly independent payoff vectors can be
achieved by changing the striking price alone.
Because of this, numerous options on the same
security are actively traded. Many of these op-
II
tions differ only in their striking prices. Adding
some of these options to our example, we get
the payoff array in table 17.
I
This system of markets is complete. To achieve
the payoff vector ($50000,$50000,$50000,0,0,0),
for example, we can transact the amounts of
the six listed call options described in table 18.
I
price 295 when the index itself is at 290 would receive
$500 x (295-290) $2500.
=
I
I
a
 a
1561 1t~d
V/H3~VII
-coolly 0141 JO Aaua~aijj’aoqi s~a~aqsaldmwxa 041 0.111311 uaaq 0.1114 AJOOtp aou030Jaad-alels lOLIC
ut ~uweaddr somamp 8uipweAo 041 Jo auo -jed 041 pui~510)JJPW 91a d WOO Jo .1.10041 041
1
suolIPAouui ~
SNOISfflEJNO~J I
1ensnun Jo uoipi~ajqoadjuaoaj amp aoj uoip?
-ueldxa .‘cpcaa r apiAojd Xew qjjm~apaAiwlaJ 51141
‘ua~oTames 041 A9 slaljavul ~ JO aaqmunu
amp JO 55031(0 um nJ spooW jo ~aqmunuamp SOAflOl ‘pp~>~’ulrnJaoun ur ul soqoJiJod .11041
]unon[nl o>tt.q salinqulic Jo suonwuiqwoo a~qtssod jatcaaW
Wu~uo1qseJut 5.10150.1W JOJ ~ciipqixa~j
041 ll~~unp~t:alaldmoa WOJJ J~JSi PIJOM 1~a~ SJ 1~nsoJ041 auole xapul 0141 SCOf) u241 aoeds
041 ui a~qe~w~r Sla)IJPW Jo w@ts1Cs amp 111141 Si JJoAed 041 Jo a~ou~ ueds Aluess030u suo~doxis
‘JaAaMoq ‘p00W C JO uon!uiJap Atau 81141 Jo UOi1 08041 JO SJJOAPd iuapuodapui Xlieoull 0141 ‘SSaj I
-vaqdmi atrnpawun uy pawnsuoo 51 WuIqlal.uos -aq1auop~-otiojI/od xapu~amp JOJ sanud a qtssod
1
4314M iv OWlS pue amu 041 :iuaJaqul paJep~S OJP 9.19141 uemp Pf.tOM 041 Jo 5011115 a~ow .114 puti
-uoo z4jvmuaou lou sainqiJuc apnpui 01 SpOOW xapui OOS d’~S’S0111 JOJ saoud o~qmssodx~surn4~
WuiuiJapaJ Aq ~TJ~T~14 -sanssi n~mouooaicuvm uo aaom Xuvmu an 0.1041 ‘asjnoo Jo ‘aopovJd ul
iqW~j8p048 810)1.1Cm atajdmoo Jo /CJOO14T 041 ooo’ocs
pul~OJOZ uaa~taqOJOI4MOWOS
anuaipnv pazi~Ct3ad5-uOuC aoj AJoeql 041 aq tsnw suoqdo ~O O1IOJTJOd 8141 Jo 1803
Jo sjdaouoo oiscq 0141 sAaAuoa ,xadvd sp.~j~ luani 111101 041 temp .10.19.1104 ‘se~idwiuoqipuoo aWpJ1
-nWic amp Jo aanlvu l~~!u4~~1 0141 Jo asnvoaq -iqoe-ou aqj -aaded 5141 Jo adoos amp puo>Caq
ssapqnop ‘000qMasla aansodxa 011111 ~ ~! Wui3lJd uortdü Al uvnb OAi13adSaJ 041 saW!1
-0.1 0.11114 soidot asaqi ‘enuvijodu.n lvo!laJoaql uoqdo qoya jo aoud 041 5J 1800 JCIIOP lC1Ol 041
I
.11041 atidsan saeaiC ~jj, ~souqeJoj aJnleJalq ~s1unOmue JCll0P tou 0.111 0.104 P108 pue iq8noq
eout?uq pur SOIuJOu030 041 ut 4DJPCSOJ Jo seam suo~dosnouCA 041 Jo SJOqUInu 041 111141 OJONJ
?‘7”~1771Ni7>: NP>’ ‘71LI~>>~’7>/~~
7 ~/ ‘7 11>.I:’>~ /7>’t>JN’14/,t/N<N////a,c*a ‘>‘//>,~~
~~<1~*fl1~’
~ 7”’”
I
N’NP 4 814<N/,Sj7t9.*//<//~~ N/~<
71
Lz:~&<t4T:1~
I
Nt #! TALJN’
\~N///7/7/N’NN/~7:/’///N’,anfla4jøsv4
RI! £tN~ ad44~~ ‘:c/N’c’7c>’
‘<V:>*717171>~>’ ,717,) “Y1<’~’<’>”7~< I
717 N, ~‘>7 7 - N/’7>
/ / / / / / >7> <N ~ ~ <>‘ / N / / : / >>71~” LI
/—N’ 77 7’> 7N’>’N J N N N <N N 2k>>, /
771
~ >/07,JNQ’/ N&N’INNSLJI N
~/.~J4”’LIN,N’ : ‘>~“~4
OQ~$”1N NO N ‘nsd
~7~6’A7 6’ 7 II
_________________________________________ N NN,7 / 7>
N/LWISNS/OOØSP NS N~!;~$ ~‘7~iN~ / >,<>~ N NS
N N
t~.~t* StL~ S at> ~fl N’ \~
I
j/ >d.,J ~,271S~ ‘3N’~>N’9 N,~ N “ NMflflP3*
‘‘N’ j”’’ $~>‘ ~‘N’>N,N, ‘/N”/’7N’N,’ ‘N’ ~ NN N N>>
/ / 77 SSSSWGAt4N N ‘N N, / /
N/N / N N N aN014*x ~ I
7 / 1 N “N N N N V /77/
N7’ <N ~ N “ N NN’
I
I
ES I
 54
tion of goods. This can be seen in the a [location
of risks in futures and options markets, the abil-
ity to refine payoffs at the racetrack, or the dis-
tribution of fat and lean between Jack Sprat
and his wife, Bubbles. The common implication
in each example is that additional markets can
improve the welfare of all concerned. That is,
given the ability to reallocate, individuals will do
so: they will exchange relatively less desirable
commodity bundles for those that are, for them,
relatively more desirable. A complete system of
markets provides this ability. Thus, an economy
~‘vith greater flexibility in production, consump-
tion and investment is uniformly preferable to
one with less.
A second recurring theme is uncertainty.
Review Questions
(1) Using the payoff array in table 3, construct
a book of bets that amounts to a pure
security for the outcome T-C-M; i. e., make a
book with the payoff vector (0,$1,0,0,0,0).
What is the net investment required for this
book of bets?
(2) Using the payoff array in table 3,, what is
the probability, implicit in the bookmaker’s
odds, that the outcome of the race will be
i’-C-M?
(3) Suppose the bookmaker offers a redundant
bet with inconsistent odds. In particular,
suppose the last bet in the payoff array in
table 1 is changed, yielding the new array in
table Qi. Construct an arbitrage portfolio (a
portfolio that shows a positive net profit in
all outcomes) from these bets.
(4) Ross shows that a necessary condition for
a collection of securities to span the state
space is that, ‘for any two states there must
be some asset whose payoffs distinguish bet-
1
ween them.” Suppose we have instead a
market in which no asset can distinguish
between two states. In particular, consider
1See Ross (1976), p. SI.
FEDERAL RESERVE BANK OF St LOI’tS
State-preference theory and the theory of com-
plete markets are one way to incorporate un-
certainty systematically in an economic model.
This is central to a theory that includes an un-
certain state of the world as a fundamental at-
tribute of a good. One result is a recognition
of the value of the ability to reallocate risk
through financial transactions. While specula-
tion in financial markets may or may not he un-
fettered gambling, the implicit transfer of risk
from hedgers to speculators still produces eco-
nomic value. ‘I’he theory of complete markets
thus provides a systematic explanation for the
popularity and value of many so-called deriva-
tive securities, such as futures and options.
II
“I I
the payoff array in table Q2, in which states
Ii
E and F are indistinguishable by the avail-
able assets. Show that this system of mar-
kets is incomplete.
(5) Consider a speculator in the cotton market
who believes that the price will rise. Assume
that storage (and leasing) of cotton is not
practical, but that a cash settlement futures
market exists. In other words, assume the
allowable transactions are those in table Q3.
What sort of payoff vector might this specu-
1
lator want? Flow could she construct it from
the available transactions? I
(6) Consider a banker facing the transactions
available in table Q4 over the next 30 days
in the international financial markets, where
S is the spot foreign exchange rate (~/$),
I
and B~and R5 are the foreign and domestic
30-day interest rates. Is this system of
markets complete? Suppose the banker
F
wants to arrange a forward exchange con-
tract of the form presented in table Q5. Ar-
range such a payoff from the available
transactions. What does the forward rate F
I.
equal? What is the intuition behind this
value for the forward rate? I
I
a
 55
I
I

I N TableQiN N N 7NNN N N N N N N~ N N N N N N,N N, NVNN

I N, PAYOFFS 8* OQtteME
/7 NNQflOt> ø~vMStI TN*CN, T~~J,4 M44NN N,M~4N Nl~~fl4N NN N

NNN1~S//MNNNN_42j7flP4?$$iNO71/NNNN,NDN/N*

N MwtñsNN NN ~ ~N’N~
sass N*0’OON
NO
ON / N0~
0 N

I ‘i4pit~çe~ $2 o sss4 “0 seei N,

NC~In$N,NVN i~NN7 -4*1 4NNN: o NNNNON ,NNNNNO~LI14N’SNH~KbGNNNN

it T4*.>Qa
N, N,
$2
N,
C’ $250
N,N, N,
$2150 N,

I $tfllRlTt
~ ,~
N,
N, SHTPt
N, ‘N ~
fN>
I SOC
A N,
N,

8>N,
*iso>oob>
C> P’”
t>~t~o~t seat
N,

s
N,

>EN’
4*S
N, N,

*N’SO~Q

zone ,za. : >,~ N N ,f r’~’o~71’ N “N N1400>060NN>, $~Ad0>‘-tiö~ N/7~i~fl0o ~Ii*_

I ~ø N S/N7N” /N>/ *071/ “ I171 >e >7
-
/ o. / N N N
$2~q4*e N,,$ N N,, ,
~
—~
snsfloa ~
$flS’r~50
505////.~4,$N’L*~’\4N’N’7Q77N/// NN’ /*N2SS ~

II a* N’N, ~G j, 6’ N,N, 71’ 4N’ $200U0GN’$L000.OdO

‘‘NNNN’
N’ N,
/~N,N
N, N’N’>
</,//,‘N,\\
N> 71
//\N’//N/N:NN’
N / N, N, N’
NN7N’N
N’ N, N,NN$N’
N,
N

N’ N ‘N N’> N, N, N, N, N,

II N, N, N, N’N, N, N, N ~“
II N,Tabte~3> N,N, ‘N’N, “’ N, N’ N,
N, N, N, ,~ N, / N, ~ ~N N, N, N,

Ir ~
IT4A14SACTIOM”>N
- N,
7/ CURSENT/,/
N, COST N, N, N N,

N~~N6’/N\
N,~\NN,Ot~NN 1NN/C~Ot/N a~L~
Set
V71*N,77N’N’PN’1<N
I ~Cetofl>71’~ 71tatet~
N’$~N~N / NN~1/N

NPaPLC
N>OPH
NN~5~ Later.
Nt~t’r’P>üQj
NJ ~F

Senns N, N, =~tN, \~NO NO NO >6’

~ P1~ 6’ N, 1QN tN N,~?LNb 0
~W~K N, N,N, ØN, >0
tantt N ~$4 N, Nft N &NNN ‘ NaNNN’ N N’ ,f N,, -A N

NN N N “N N “ N N N N N N N, N N N,
N N N N, NN N NN N N N, N N, N N

INN N N NN NN N NNN,\ NNNNNN, N NN N N N N N

II N 7N NN
N

N N
N

N
N N NNNN/NN

NN N N N PAV0FP~NT*COHTSACTS~
N

N
NN

NN
NN

N
N

N
N N

II
IN~NLIN

N N N N NN N N N N N N N N

N N N PAYOFFS TO CONTRACTS

I N $uy orward
CONTRACT $ NOW $0 N $ now
0 ‘$ F
*Oday S —$1
30-4ay

MARCH/APRIL 1991
a
 56
REFERENCES
Arditti, Fred D., and John Kose. Spanning the State Space
with Options,” Journal of Financial and Quantitative Analysis
(March 1980), pp. 1-9.
Arrow, Kenneth J. The Role of Securities in the Optimal
Allocation of Risk Bearing,” Review of Economic Studies
(April 1964), pp. 91-96.
_______ “Insurance, Risk and Resource Allocation,” in
Essays in the Theory of Risk-Bearing (Markham, 1971),
pp. 134-43.
Arrow, Kenneth J., and Gerard Debreu. Existence of an
Equilibrium for a Competitive Economy,” Econometrica
(1954). pp. 265-90.
Asch, Peter, Burton G. Malkiel, and Richard E. Ouandt.
“Market Efficiency in Racetrack Betting,” Journal of
Business (April 1984), pp. 165-75.
Chicago Mercantile Exchange. International Monetary
Market Yearbook (1973-74).
_______ Index and Option Market Yearbook 1982 (CME
Statistical Dept., 1983).
______ 1989 Statistical Yearbook, Futures.
Copeland, Thomas E., and J. Fred Weston. Financial Theow
and Corporate Policy (Addison-Wesley, 1983).
Cox, John C., and Mark Rubinstein. Options Markets
(Prentice-Hall, 1985).
Daily Racing Form (Midwestern Edition) (News America
Publications, July 20, 1990).
Debreu, Gerard. Theory of Value: An Axiomatic Analysis of
Economic Equilibrium (Cowles Foundation for Research in
Economics at Yale University, 1959).
_______ “Existence of Competitive Equilibrium,” in Hand-
- Townsend, Robert M. “On the Optimality of Forward
book of Mathematical Economics, Volume II (Amsterdam:
North-Holland, 1982), pp. 697-743.
Friesen, Peter H. “The Arrow-Debreu Model Extended to
Financial Markets,” Econometrica (May 1979),
PP~689-707.
Gabriel, Paul E., and James R. Marsden. “An Examination
of Market Efficiency in British Racetrack Betting,” Journal
of Political Economy (August 1990), pp. 874-85.
FEDERAL RESERVE BANK OF St LOUIS
II
Geanakoplos, John. “Arrow-Debreu Model of General Equil-
ibrium,” in J. Eatwell, M. Milgate, and P. Newman, eds.,
The New Pa/grave, A Dictionary of Economics, Volume I
(London: MacMillan Press, 1987), pp. 116-24,
. “An Introduction to General Equilibrium With In-
complete Asset Markets,” Journal of Mathematical Econom-
ics (1990), pp. 1-38.
Hirshleifer. Jack. Time, Uncertainty, and Information
(Oxford: Basil Blackwell, 1989).
Hirshleifer, Jack, and John G. Riley. “The Analytics of
Uncertainty and Information — An Expository Survey,”
Journal of Economic Literature (December 1979),
pp. 1375-421.
McKenzie, Lionel, “On Equilibrium in Graham’s Model of
World Trade and Other Competitive Systems,” Econom-
etrica (1954), pp. 147-61.
Mossin, Jan. The Economic Efficiency of Financial Markets
(Lexington Books, 1977).
Myers, Stewart C. “A Time-State-Preference Model of
Security Valuation,” Journal of Financial and Quantitative
Analysis (March 1968), pp. 1-33.
Radner, Roy. “Equilibrium under Uncertainty,” in Handbook
of Mathematical Economics, Volume II (Amsterdam: North-
Holland, 1982), pp. 923-1006.
Ross, Stephen A. “Options and Efficiency,” Quarterly Journal
of Economics (February 1976), pp. 75-89.
Rubinstein, Mark. “Securities Market Etficiency in an
Arrow-Debreu Economy,” American Economic Review
(December 1975), pp. 812-24.
Stoll, Hans R., and Robert E. Whaley. “The New Option
Markets,” in A.E. Peck, ed., Futures Markets: Their
Economic Ro/e (American Enterprise Institute for Public
Policy Research, 1985).
Markets,” American Economic Review (March 1978),
pp. 54-66.
“Goldman Sachs Starts Selling New Warrants Betting Yen
vs. Mark,” Wall Street Journal, November I, 1990.
Wilson, Charles. “Incomplete Markets,” in J. Eatwell, M.
Milgate, and P. Newman, eds., The New Pa/grave, A Dic-
tionary of Economics, Volume II (London: MacMillan Press,
1987), pp. 759-61.
II
U
a
 57

Answers to the Review Questions

(1)Try:(30C,—G0C,20C,--25c,1000,—85C). there are six states and five non-redundant

The net investment is 20C. securities—the system of markets is

(2) Odds are given by: incomplete.

= I C 1.00—20
.20 = 4 = 4-to-I odds. (,5)She
in>1
Z in the
would
table Al.want
This to purchase
could
high-price state, as
a payoff
described
be obtained by the
I = 4 P(e) = -~ = 20%. transactions listed in table A2.

P(e) (WYes, it is complete. Combine the trans-

actions listed in table A3. Thus, F =

(3)Try: ($240) —$240,$160, —$2204400, —$320)- S(1 + R~)1(1 + R5). This is simply a statement
(4) Compare: (0,0,0,0,1,0) with: (0,0,0,0,0,1), of the covered interest parity condition,
where the elements of the vectors repre- F(1 + R5) S(1 + R ,~),which prevents arbi-
=

sent the numbers of each option to buy or trage between money markets in different

IF sell. The last two options are redundant:

/ // / N’,, / /
countries.
N, / ,,/N’N~/N> N’ N’ ~ 71 //

II / / // /71/ ,>> ‘~~// ,N’~NN’ / ‘N” / 71 N’NN, ;/ /

U / N’T N’ ‘N’ /N,,, N~ N N”~’~iflTtaA$s*Fs /

/7/7,/N. :Mctart~ca*c ~,e4<cOnO~\r StGt~/k aes tNC*N”f /~N~ / 17171’

“NS 71f tser~,J71’uSr 4 tatet~ ~ tar
11 ~I
N’7>’
‘>

NNN///!!//
N’ N’
// /

/‘/NN//
/~

7N
‘N:

I”/Mx/N,/”
/N N,NN/ /
4/
,//
’’A ///
~
/N,N/’
/
~w~<j~q~&
-‘‘ , /‘///
“7,,
N>
~
N / N’//N/
~

N
<NN
N’>
N’/
// //

IN IN~’>
— ‘7 > / / N, /‘N / ~N, ~N // “N,~ N’ / “N’ NN NN’

I // “ N> / N’ N,> N, N /

/ k/N, / N, / // /N~ N” / ‘~1 N’ N,

>N >N’~&~ N / >~N’>~ N, NN71/N N~/N’ N,

/ “‘N / ‘N N, N, N’,~ ~ ,,,N, N, N,~

“ ~ N, ‘N “N’ N, *T**CLMMflY~! N’ / N’N’

I N,// // N’N’NN ~ t
71tasbI~ CMtbkNN’ NNNNN’C~

N’NN’>, N’ N, tNQSTN’ 3 140* i~Met ~ Latec/ t~*

I NNNI771
NN, laN’dN’~N’~~t~frpN’P
N,N>~N,N,N\N>/N’\N’N’~N~NN/N~N’\ ~ N,f/$711
INN’71 ,~ / ~NPj71
N9
N,N,~
7’
f~> ~ j NON,
‘tIN/N ~P~rL PtC
:!L~p~c ~RC’~~~>
NSNPtNNN>N /~tj4,NN N N~“ ‘N” N

IN
I :,:IN.
N,N
N,NN

N,
“44~,>,,N’

/
OthIdS~ab~tame/N’N’NNN

N, N
N,N>f~P~~O4tR~NN,•N~

N, N,~s1 N
\\N

N,
~
NN
N,\

/NNNNN.N1

NN,
* N,~~N’

NN~

N,
~NNNI

~ Nfl’
N, t(FW~UN
t/IN/ IIN N
“N,

N, N N \N,/,\ N I>NN~NINN’71 ‘N NN71 N N\NN’NN, N\’N> NN> N NN,

N, N,N N, N, N,

I Tab1a~ N,

CONThACY
N N/N, NNNN

$ NOW
N’ N N, N NN,,~fltPT~C>~S
~ ~‘rj
day $ ~
N ~NN’N NN

ctntt’&Iesv 1+R N,

N71N erroWi N ~ N,NG N N NN QNN NN _~NN N~ N~ N

I Net:
N “~
0
NN
0
NNN’N~N ,,1
F
N I
—$1
N

I
MARCH/APRIL 1991